[PLUG] Palimpsest v. e2fsck v. ?

[Daniel Pittman]

On Tue, Dec 28, 2010 at 18:02, John Jason Jordan <johnxj at comcast.net> wrote:

> Having just finished various dist-upgrades of my desktop from Intrepid
> to Lucid, and healed a sick RAID in the process, after booting to Lucid
> I am presented with a warning that disk failure is imminent. The disk
> in question is not part of a RAID; it is a 1 TB disk that I added
> separately. The disk has always been empty, so there is no worry about
> losing stuff.
>
> The warning comes from Palimpsest, the new GUI disk utility that comes
> with Lucid. I launched Palimpsest as root so I could use its GUI to see
> if I could fix the drive. After unmounting it I clicked on the Check
> Filesystem button. Palimpsest then reported in a popup window
> that the disk was clean, while the main window still insisted that disk
> failure was imminent. According to the Smart Data window the
> Reallocated Sector Count is in red and labeled as failing, also the
> Airflow Temperature, which it reports as 49C.
>
> While wondering about these reports I reformatted the drive as ext4 (the
> Lucid default, evidently). The reformat went fine without errors.

The answer to that is simple: the "Reallocated Sector Count" is the
number of failed sectors on the disk surface that the firmware has
replaced with good sectors from the pool of spare space it uses.  The
airflow temperature is, as you would imagine, the highest temperature
the disk has run at.

Both of those are properties of the physical hardware, not of the
content - so checking the file system does nothing about those
underlying faults.

> At that point I concluded that Palimpsest was given to prevarication.

Nope.  It is honestly telling you that this disk has run hotter than
the manufacturer thinks is safe, and that it is showing significant
problems with the disk surface.

> But just to be sure I decided to run e2fsck on the disk. This gave me
> a strange error message:
>
> Superblock invalid, trying backup blocks ...
> Bad magic number in super-block while trying to open /dev/sdc
> The superblock could not be read or does not describe a correct ext2
> filesystem. If the device is valid and it really contains an ext2
> filesystem (and not swap or ufs or something else), then the superblock
> is corrupt, and you might try running e2fsck with an alternate
> superblock:
>        e2fsck -b 8193 <device>
>
> First, note that e2fsck thinks it is looking for an ext2 filesystem. I
> read all the way through the e2fsck man page looking for an option to
> specify a particular filesystem, but couldn't find it.

ext2, 3, and 4 are occasionally referred to as ext2 by the tools
because they all descend directly from each other, so are close enough
that the code is common.  Don't worry about that tiny mislabelling.

> Second, "8193" is the number to use if the block size is 1K, 16384 for
> filesystems with 2K block sizes, and 32768 for 4K block sizes. So
> before I try to fix the superblock I need to figure out what the block
> size is. The Palimpsest GUI doesn't say, and I don't know how to find
> out from the command line.

My magic says '4K', because everything gets formatted that size these
days.  Theoretically you can use tune2fs to ask ... but that needs the
superblock.  Your best bet is to throw away the faulty disk ^W^W^W^W
try a non-destructive fsck with 4K, and then try 1K if that doesn't
work.

> This is not an urgent matter, but if the drive is really failing I'd
> like to know so I can return it to Maxtor for replacement.

It absolutely, without question is: every increment of that
"Reallocated Sector Count" is another bad sector on the disk.  (My
guess is that your fsck failure was another bad sector, unluckily
placed, and another format would "fix" it for a while, but there are
other causes.)

Anyway, the disk is really dying, and the software format is way too
high level to ever come close to touching the root of the problems. :)

Regards,
    Daniel
-- 
✣ Daniel Pittman            ✉ daniel at rimspace.net            ☎ +61 401 155 707
              ♽ made with 100 percent post-consumer electrons